Regression plot for the calibration data at Christchurch. Badgecalibration data with quadratic ﬁts through the origin (10 min means,with normal incident data removed).
were similar to those on the calibration days. The estimated 2
uncertainty, relative to the reference instrument, resulting fromthis approximation in eqn. (1) is
3.2 Temperature dependence
Possible temperature dependences of the UV monitoring badgewere investigated in two ways.Firstly, during the calibration periods, we recorded theambient temperatures at 10 min intervals at the calibrationsite. The ranges of temperature during all of the badges mea-surements discussed are shown in Table 2, which also includesinformation about the ozone amounts and minimum SZA onthe observation days. In the regression analysis we included aterm to allow for temperature-dependence in the badge output.Becausethetemperaturerangewasrelativelysmall,wewereabletoconﬁrmonlythatthetemperaturecoefﬁcientmustbelessthan0.5% per
C.Because the temperature difference between calibration con-ditions and mountain conditions is larger than the range of temperatures available for calibrations, we carried out a furthercheck of the temperature stability. Under clear sky conditionsat Christchurch on 9 June 2004 from 12.20 to 12.40 NZST,the UV monitoring badge, which had been stored in a fridge atapproximately8
Cfortwohours,wasplacedonahotplatewitha large thermal mass, pre-heated to approximately 80
C, whichwas then placed in the sun and allowed to return to ambienttemperature. By heating the whole badge in this manner weinclude any temperature coefﬁcient of the electronics as well asthe sensor. Model calculations showed that any change in UVintensity over the period of measurement was less than 1%. Theresults were as follows.Ambientbadgereading(fullscale1024counts):73
2countsBadge reading after 10 min on the hot plate: 78
2 counts.Thus, even with this extreme change in temperature, thechange in output was relatively small: 7
3%. The ambienttemperature at the observation time varied from 9.5
C. If the badge reached 80
C, the temperature coefﬁcientwould have been
C. In reality, the real equilibriumtemperature would be intermediate between the ambient airtemperature and the plate temperature. Based on these resultswe estimate that the temperature coefﬁcient should be less than0.3% per
C.The temperatures at Mt Hutt are at most 15
C lower thanduring the calibration period. Based on the above sensitivitytests, we estimate that the observations at Mt Hutt may beunderestimated by up to 5%.
3.3 Cosine response
The angular response of the UV monitoring badge was mea-suredat Lauderusingaquartz-halogenlamp at adistanceof 1.3m. We found that it showed a reasonable approximation to thecosine response. For incident angles up to 30
, it underestimatesthe true cosine response by up to 5%. The error is largest at 60
,where it underestimates the true response by 12%. For largerangles the error reduces and at 80
the error is less than 5%.Although these response ﬁgures do not represent the state of theart, they are comparable with those of several research gradeUV instruments,
and they are signiﬁcantly better than somecommercially-availableUVdetectors.Inthepresentapplication,where the instrument orientation is non-horizontal in a snow-covered, mountainous area, the effects of any errors in cosineresponse become more important.
3.4 Directional effects
Unlike previous studies of the effects of altitude and snow cover,the present study investigates the effects on personal exposuresfor a sensor located on the skier’s lapel. Individual UV badgereadings, which have arbitrary orientation, will be designated asUVI
. In this case, the peak UVI readings correspond to timeswhen the badge orientation is normal to the sun. These will bedesignated UVI
.During calibrations at Christchurch, the dosimeter badgewas generally positioned to measure the irradiance falling ona horizontal surface. To quantify the difference between UVI ona horizontal surface with that normal to the sun, the badge wasoccasionally re-positioned to point directly towards the sun onthe16October,resultinginhigherdatapoints,asshowninFig.2.Theratio(UVI
/UVI)oferythemalUVradiationonasurfacenormal to the sun compared with that on a horizontal surfacehas been measured previously at Lauder,
and the results areshown in Fig. 3, along with the corresponding measurementsfrom the calibration sequences at Christchurch.
Minimum SZA, ozone amounts, and temperatures during all measurement periods with the UV monitoring badge. For Mt Hutt, the ﬁrstvalue is for the base altitude, and the second value (in parentheses) is for the top of the ski-ﬁeld. Ozone values were from the NASA E-P TOMSsatellite instrumentTemperature at selected times (NZST)/
CDate (and day), 2003 Site SZA
Ozone/DU 09:00 12:00 15:0015 Sept (258) calibration 1 Christchurch 47.1 364 6.5 12.9 14.112 Sept (255) comparison 1 Christchurch 48.3 361Mt Hutt 48.3 345
4.1) 6.7 (
4.3) 4.2 (
3.7)15 Oct (288) comparison 2 Christchurch 35.5 356Mt Hutt 35.5 353 2.6 (1.8) 9.9 (4.4) 9.4 (5.8)16 Oct (289) calibration 2 Christchurch 35.2 379 10.9 14.1 15.117 Oct (290) calibration 3 Christchurch 34.8 385 16.6 14.6 13.6
P h o t o c h e m . P h o t o b i o l . S c i .
, 2 0 0 5 ,
, 4 2 9 – 4 3 7
4 3 1